When Building Gets Cheap, Decisions Matter More, with Ashley Roach

WEBVTT
00:00:00.000 --> 00:00:13.199
When fast prototyping makes it possible to turn an idea into something testable, whether that's a software tool, a 3D-printed object, or a new product concept, the constraint changes.
00:00:14.210 --> 00:00:17.050
The challenge is no longer whether we can build.
00:00:17.410 --> 00:00:25.379
It's deciding what to build, what to learn, and which decisions need more care before they become difficult to reverse.
00:00:26.899 --> 00:00:30.399
That acceleration can change the moment before commitment.
00:00:30.949 --> 00:00:41.859
When building feels inexpensive, jumping into the build can feel like the fastest way to reduce uncertainty, even before we've defined what we need to learn.
00:00:43.310 --> 00:00:49.719
Speed can make the building process exciting, but it can also make tweaking feel like progress.
00:00:50.920 --> 00:01:03.890
In this conversation, Ashley Roach and I talk about how to keep fast prototyping connected to customer feedback and real value, whether you're working independently or inside a larger organization.
00:01:05.060 --> 00:01:13.799
We discuss the build trap, the difference between one-way and two-way doors, and how more people across a company may become builders.
00:01:14.989 --> 00:01:22.890
We also look at how AI can support that work without replacing the judgment needed to decide what deserves to move forward.
00:01:24.450 --> 00:01:28.250
I'll tell you more about Ashley Roach after this brief introduction.
00:02:14.503 --> 00:02:21.574
Ashley Roach is a product leader and builder who has spent his career turning early ideas into products people use.
00:02:22.413 --> 00:02:30.014
He was part of the founding team behind Cisco DevNet, which grew into a community of more than five hundred thousand developers.
00:02:30.993 --> 00:02:43.493
He later led product for Learn with Cisco, overseeing its applications, user experience, product marketing, and commercial offers across Cisco U and Networking Academy.
00:02:44.653 --> 00:02:54.683
Today, Ashley builds independently and writes Beyond the Prototype, a Substack about what one person can realistically ship, operate, and monetize with AI.
00:02:55.567 --> 00:03:07.530
Let's dive into the interview You've spent your career on early uncertain product problems, and you write about building real products with AI.
00:03:08.449 --> 00:03:17.710
Once AI and these other fast prototyping methods made building cheap and fast, what did you find became the hard part?
00:03:20.210 --> 00:03:22.490
I think that one of the...
00:03:22.780 --> 00:03:26.400
one of the difficulties, of course, is what to build in the first place.
00:03:27.020 --> 00:03:37.370
So I think the desire to hold yourself back is a lot, maybe it's a lot different now.
00:03:37.370 --> 00:03:39.580
You don't have to hold yourself back as much.
00:03:40.080 --> 00:04:04.831
I found that in the past where I have kinda had a, a governor in my mental model to say,"Oh, that's too expensive to build," or,"That's too m- too risky." Now I think the time in which you can experience that creation process to testing something is so short and inexpensive, frankly, to be...
00:04:04.842 --> 00:04:17.692
If you're building with AI specifically, that you can try a lot of different things and that really makes a, like a, just a lot of difference between what was now versus, say, two years ago
00:04:19.660 --> 00:04:26.949
And then I think producing things really quickly makes you more excited about what it is that you're developing.
00:04:28.129 --> 00:04:31.043
And so you just kind of wanna keep going faster and faster
00:04:32.189 --> 00:04:32.848
Yes.
00:04:32.858 --> 00:04:46.149
One, one funny thing though is I had an interchange with the AI tool that I'm using because I've been debating about, hey, when do I add functionality to add billing, right?
00:04:46.158 --> 00:04:47.699
How do I make money off of this thing?
00:04:48.869 --> 00:04:51.449
And it kinda called my bluff.
00:04:51.449 --> 00:04:59.309
It's like,"Hey, you're really good at shipping things, but you don't have any proof this, this system is valuable to customers.
00:04:59.879 --> 00:05:02.829
You don't have enough people testing it." And so it was a really...
00:05:03.459 --> 00:05:04.278
I was like,"Oh, man.
00:05:05.269 --> 00:05:38.959
It's right." And so there is also that element of because it's so easy to build things, and I love buil- that building process, that it's easy to get like stuck in that loop to say,"Oh, I could tweak this little thing," or,"Oh, I could tweak that little thing." And so that, that is something that I- I've had to dial back, my, my desire to, to tweak versus how do I get this tool in front of people to give me feedback to know whether it's worth building even more?
00:05:40.322 --> 00:05:51.853
So in that project, the AI actually pushed back against you saying,"Hold on." Now, was that because you pre-trained it with anything specific from your past experiences or anything?
00:05:53.533 --> 00:05:54.392
Not really.
00:05:54.442 --> 00:05:58.853
I'm using a combination of the main frontier model companies.
00:05:58.862 --> 00:06:04.632
So Claude from Anthropic and then Codex and ChatGPT from OpenAI.
00:06:05.853 --> 00:06:06.262
I...
00:06:06.452 --> 00:06:13.362
While it has some memory, to be honest, I didn't use Claude very much up until I started this project.
00:06:13.362 --> 00:06:16.692
So it didn't have, it didn't have a lot of memory about me.
00:06:16.723 --> 00:06:27.302
And yeah, it just took that I guess you could call it an antagonistic stance to just help me think through what the parameters are of success.
00:06:27.432 --> 00:06:31.473
And I may have asked it like,"Hey, I wanna do this thing.
00:06:31.473 --> 00:06:37.692
Is this the right time to, to do it?" And so I set it up with that context.
00:06:38.242 --> 00:06:47.072
But beyond that it really just acted like a thought partner at that point or a co-founder, and and so that was surprising and also very helpful
00:06:48.716 --> 00:06:52.406
I was just talking to a neighbor who has a 3D printer.
00:06:52.797 --> 00:06:56.367
I'm hoping to talk with a few more people'cause my family would like to get one.
00:06:56.706 --> 00:07:02.723
But he's producing all sorts of products with this fast prototyping machine.
00:07:03.483 --> 00:07:11.990
Now he says he, he goes to an online place to download the specs and he's been printing things.
00:07:11.990 --> 00:07:14.399
But then the whole business thing started to come into place.
00:07:14.439 --> 00:07:19.089
"What if I would make this for people and have an online shop?
00:07:19.089 --> 00:07:39.387
And I could start with something small like earrings." And so a lot of the things that he was talking about was,"Hey, I can produce this and just put it out into the market." But he's on his own with his fast prototyping machine, and he doesn't have that co-founder antithesis fighting him on what do the customers really want?
00:07:39.387 --> 00:07:41.576
Are you sure this is something you really want to do?"
00:07:43.872 --> 00:07:43.983
Oh, yeah.
00:07:44.033 --> 00:07:49.072
I actually have a funny, similar example that's non-technical at all.
00:07:49.142 --> 00:07:53.593
I started doing leatherworking at about a year ago.
00:07:54.062 --> 00:08:01.552
And, one of my other sort of entrepreneurial ideas was maybe I could just make some of these wallets and put them up on eBay.
00:08:01.632 --> 00:08:02.122
Or Etsy.
00:08:02.812 --> 00:08:15.242
And so I went through an analysis process on that idea, and it did a whole bunch of margin calculation there's a book called The Lean Startup, which is pretty well-known at this point.
00:08:15.333 --> 00:08:27.642
But it talks a lot about what's-- what are good tests that you can run that don't cost a lot of money if any, that can then prove whether or not there's a market or demand for the thing that you're creating.
00:08:28.242 --> 00:08:36.903
And so even within that context of the wallets, we went back and forth about should I do a wallet?
00:08:37.293 --> 00:08:41.312
Should you do something like a purse or a bag?
00:08:41.893 --> 00:08:45.163
'Cause that's higher margin, and people will pay more for it, like your unit.
00:08:45.222 --> 00:08:48.192
And did all the unit cost analysis, all that kinda stuff.
00:08:48.842 --> 00:08:51.003
So it's pretty fascinating.
00:08:51.113 --> 00:08:58.202
The, the other thing I will say, just kinda circling back to your question about, hey, did I tee any of this up with Claude.
00:08:59.413 --> 00:09:13.243
I have tried doing some things that might also be helpful for the audience, which is I've been using a tool called Hermes Agent, which there's a thing called Open Claw that a lot of people have heard of.
00:09:13.253 --> 00:09:20.783
Hermes Agent and Open Claw are the two main open source I guess agentic agentic tools.
00:09:20.863 --> 00:09:34.793
And what I created was my own venture studio, call it, within Hermes Agent, and it's really just a set of markdown files that define these different personas of what the agent is supposed to behave like.
00:09:35.302 --> 00:09:45.192
And for example, I have one that's a chief of staff agent, and the other one is a researcher agent, and the other one's a writer agent, and then there's like a CFO agent.
00:09:45.692 --> 00:10:07.722
And so I can toss an idea and say,"Hey, I really wanna do this, wallet thing," or if I wanna do the 3D-printed parts bus- business like your friend or neighbor, then you can run those ideas through something like that, and it does provide a little more structure than,"Oh, hey, I got a randomly helpful result from Claude," or, or OpenAI.
00:10:10.335 --> 00:10:27.394
Now I know we're talking about like the leatherworking, the 3D printing, the software apps but this also scales up to the bigger corporate environment where there are teams of people.
00:10:27.735 --> 00:10:31.684
And so you used Hermes to create, the agents.
00:10:31.855 --> 00:10:33.315
Did you say that you created agents?
00:10:33.315 --> 00:10:33.595
Yeah.
00:10:33.804 --> 00:10:43.745
So you created agents to represent those functions, but those are functions that would be normally operating in a big company, right?
00:10:44.174 --> 00:10:49.524
So you knew who you needed and who you need to talk to get what information.
00:10:50.254 --> 00:11:01.804
So I just wanted to check with you that is it the same kind of thinking that you're applying to your own solo adventures as to what you did in the big corporate ventures?
00:11:04.355 --> 00:11:07.995
I'd say yes, but with some obvious caveats.
00:11:08.174 --> 00:11:19.914
The big company, all the different functions if you don't, if you don't work at a large company or a large organization or haven't, then it might be difficult to know what those are.
00:11:20.475 --> 00:11:21.054
Those...
00:11:21.585 --> 00:11:33.981
Again, though, you can turn to AI and say"Hey, I wanna create a, a mini business here to analyze opportunities," and it can guide you through that sort of thing.
00:11:34.461 --> 00:11:49.471
There's another tool called Paperclip, they're trying to create literally a corporate structure with given budgets to each of these agents so that in theory you can say,"Hey, I have this problem.
00:11:49.751 --> 00:11:56.351
Go solve it." And those agents will go, and they'll run up against budgets, and then they'll produce you a result.
00:11:56.932 --> 00:11:59.721
I think they're...
00:11:59.861 --> 00:12:20.302
with big organizations, a lot of the complexity comes down to things like change management, stakeholder management, as opposed to,"Hey, I can build a thing." A lot of cases, the I can build a thing, you get stopped as fast as you created it because it's no, that's not in the roadmap.
00:12:20.302 --> 00:12:23.121
Or, no we don't have time to do that.
00:12:23.131 --> 00:12:24.331
That's a distraction.
00:12:24.881 --> 00:12:29.761
And, uh, so I think in a large organization, it can...
00:12:30.381 --> 00:12:39.121
These tools can help you accelerate maybe the thinking process and pre-presenting something that fits within that context.
00:12:39.121 --> 00:13:04.652
But you still have to go through the, you know, the stakeholder meetings, getting approvals, um, governance, all of the, all of those words that, you know, someone who's got entrepreneurial spirit and drive run up against and go,"Oh, man, this isn't very fun." Where in the-- in your own world when you're doing these things you're maybe your worst enemy'cause you don't run up against those.
00:13:04.652 --> 00:13:16.601
And you're like,"Hey, I can just blindly go walk off this cliff." And big companies are kinda designed to manage risk and not help have projects walk off cliffs too easily.
00:13:17.172 --> 00:13:22.404
So I think those are some of the, the differences, at least from my observation at this point.
00:13:24.248 --> 00:13:31.288
you mentioned paperclip applying different costs to different functions for an entrepreneurial project.
00:13:32.248 --> 00:13:35.378
But then there's also the cost centers for the big corporation.
00:13:35.447 --> 00:13:45.638
And like you said, they're designed to manage risks and slow things down to make sure that the decisions are appropriate for the business, right?
00:13:45.850 --> 00:13:46.500
Mhm, mhm.
00:13:47.748 --> 00:13:55.801
So you wrote that the implementation cost and the decision cost are not necessarily the same thing.
00:13:56.701 --> 00:14:00.941
So can you explain a little bit more about what you mean by that?
00:14:02.791 --> 00:14:09.717
Yeah, I think you might be referring to the article that I wrote about kind of one-way doors and two-way doors.
00:14:09.727 --> 00:14:12.148
I think that's probably what you're getting at.
00:14:12.618 --> 00:14:25.268
So for those of you who aren't familiar with the one-way door, two-way door concept, this was something that came from Jeff Bezos back, I guess, when he was early on in Amazon.
00:14:26.097 --> 00:14:48.587
And what he had observed was companies and organizations tend to spend maybe the same amount of time on a decision, and thoroughness on a decision that is what he considers a two-way door, so it's reversible as you do with a one-way door, which is not reversible.
00:14:49.197 --> 00:14:58.018
And so the heuristic that he presents is for decisions that are these two-way doors, don't dither on them too long.
00:14:58.248 --> 00:15:04.548
Just make the decision and try it and see, because you can always walk it back basically.
00:15:05.437 --> 00:15:11.572
Where you have one-way door situations, that's where the thoroughness and deep deliberation is more necessary.
00:15:12.123 --> 00:15:15.513
And so that you know, that is kind of the framework itself.
00:15:17.283 --> 00:15:21.482
I guess hanging above the door is the decision cost.
00:15:22.322 --> 00:15:37.533
Decision cost when you're balancing these fast prototype, fast turnaround, quick tweaks to make products, but you have these decision doors and there's a real cost associated with them.
00:15:38.212 --> 00:15:46.023
So just how do you think about the cost of those decisions when you can implement so fast?
00:15:48.945 --> 00:15:54.284
And you're, are you thinking in terms of the big organization or the small situation that I'm in?
00:15:55.489 --> 00:16:05.178
I would say the small situation you're in, but, and then can it be scaled to the larger organization, to people that are working in a larger organization?
00:16:06.455 --> 00:16:06.725
Yeah.
00:16:06.725 --> 00:16:17.105
I th- I think this, this basically falls into the larger category of, hey, how do you do innovation at large or small companies?
00:16:17.764 --> 00:16:21.894
And so The, the way...
00:16:22.335 --> 00:16:30.934
One of the, one of the f- again'cause I'm a product person frameworks are gonna come up often, yeah like a lot of us listening here.
00:16:31.024 --> 00:16:39.544
So the, the framework that I got enamored with, and I still think it's interesting is called the Zone To Win.
00:16:40.054 --> 00:16:43.424
So Geoffrey Moore's famous strategy person.
00:16:43.975 --> 00:16:58.634
And he had this idea of Zone To Win, which ties to another concept of this crossing the chasm problem that small companies evolving or small products evolving into larger ones.
00:16:59.524 --> 00:17:03.215
Kinda how do you get from your early adopters to your mass market?
00:17:03.884 --> 00:17:14.055
But in Zone To Win just switching back to that, he divides essentially innovation into one of those quadrants.
00:17:14.704 --> 00:17:41.603
And the, the idea that he puts out there is you actually need a completely different management and operating mode for those projects or those groups that are doing innovation compared to these other three quadrants, which are like productivity quadrant, where you're essentially have a cash cow and you're trying to, make it as profitable as possible and so on.
00:17:41.979 --> 00:17:49.588
So that kind of moves all the important decision points for innovation to one door, right?
00:17:49.588 --> 00:18:03.749
So pe- people are innovating and coming up with ideas but it still has to pass through a door in order for it to be accepted for development for, for customers to actually use
00:18:05.268 --> 00:18:09.817
the way I think about it is it's not so much like those doors are governance gates.
00:18:10.498 --> 00:18:15.468
It's more the, is this something that's an irreversible decision?
00:18:16.317 --> 00:18:21.607
So for example, like I'll use from my experience just recently, at Cisco.
00:18:22.087 --> 00:18:26.617
When we were designing and launching Cisco U we kept it...
00:18:26.667 --> 00:18:28.367
we definitely kept it separate.
00:18:28.428 --> 00:18:34.637
We kept it on an island because we were trying to follow this, this principle of something similar to zone to win.
00:18:34.798 --> 00:18:44.019
And we didn't we didn't really trouble ourselves so much with a lot of one-way doors when we were working in the project itself.
00:18:44.108 --> 00:18:47.298
In fact, we're like,"Hey, do we need a recommendation thing?
00:18:47.308 --> 00:18:58.638
Do we need to embed AI in it?" We were, we were having a lot of fun coming up with ideas to make a brand new and attractive engaging learning experience.
00:18:59.239 --> 00:19:06.848
And so there were a lot of, say, two-way doors in that mode because we didn't really have anyone except for just testers.
00:19:06.858 --> 00:19:14.429
So it was like low risk types of decisions as we were going through and trying to find product market fit, essentially.
00:19:15.009 --> 00:19:25.469
But when we got to the point of doing things like,"Hey, we need to implement entitlements and billing and all those things," those were more one-way door decisions.
00:19:25.469 --> 00:19:38.648
And so we spent more time thinking about,"Hey, is this the right way that we're gonna display content?" Is it if you're a, if you're a free account, do you get to see everything, but you can't click on stuff?
00:19:38.679 --> 00:19:42.538
Or do you just do you just see what you have access to?
00:19:42.608 --> 00:19:48.108
Because the previous system had worked like that, where you only got to see the stuff that you were entitled to.
00:19:48.628 --> 00:20:01.459
So we were thinking,"Hey, maybe we'll use some product-led growth ideas, and if someone can see what they don't have access to, they'll they'll click through and purchase, make a purchase." So I think those...
00:20:02.449 --> 00:20:13.148
there are those bigger, maybe architectural or structural issues that are the one-way doors, and even within an innovative environment.
00:20:14.259 --> 00:20:17.679
And then there's maybe a lot more two-way door ones.
00:20:17.689 --> 00:20:24.028
So that's maybe the clarification I would throw out there and see if you think that makes sense.
00:20:25.655 --> 00:20:26.935
Yes, I think so.
00:20:27.256 --> 00:20:38.017
So the Doors aren't decision gates, but it's, um, it's the type of decision that you're looking at and it depends on the question, the problem that you're trying to answer.
00:20:39.220 --> 00:20:39.509
Yeah.
00:20:39.509 --> 00:20:53.349
It's a, it's a mirror that you use to say,"Hey, is this thing I'm, like, spending hours and hours trying to figure out something that I can just ship and get that feedback?" Because it's a two-way door thing, we can always pull it back.
00:20:53.950 --> 00:21:01.930
Or is it more of a long-term systemic type of decision which is gonna need that deliberation?
00:21:03.180 --> 00:21:11.087
From, I'm thinking of an engineering team making decisions about a product on paper.
00:21:11.900 --> 00:21:17.430
And while it's on paper and doing prototypes, you can, answer your questions.
00:21:17.910 --> 00:21:19.569
A lot of that's reversible.
00:21:20.251 --> 00:21:20.342
Uh-huh
00:21:20.392 --> 00:21:27.092
But in your example, like e- even within that, there are some decisions that are not reversible.
00:21:28.051 --> 00:21:50.241
So it really takes, I guess, an awareness to look at the decision that you're making and actually pause and just evaluate if it has, if it is irreversible, if you lock this in now, it's going to, lock in a lot of other features or characteristics about what you're designing later.
00:21:50.342 --> 00:21:51.082
Is that right?
00:21:53.582 --> 00:21:53.701
Yep.
00:21:53.711 --> 00:21:54.172
Yeah, that's right.
00:21:54.172 --> 00:21:54.392
Yeah.
00:21:54.392 --> 00:21:58.031
And there, I'll throw another,'cause I lo- I love all these different books.
00:21:58.051 --> 00:22:12.201
I'll throw another book out there that goes into these, I think some of these concepts a little more which is called Working Backwards, and it was written by, I think, a couple of senior leaders from Amazon.
00:22:12.662 --> 00:22:25.342
And it talks about how Amazon operates in general, but I believe there's a use case they use Kindle as a use case to go through a few of their processes.
00:22:25.791 --> 00:22:31.602
But but I think the one-way door, two-way door thing comes up as, as part of that as well.
00:22:32.315 --> 00:22:45.347
you've worked with teams of people to develop a product, And I would imagine that it- it's not always the project manager that is looking at decisions in this way, al- although primarily probably.
00:22:46.258 --> 00:23:04.167
But there could be things that the senior design engineer or the electrical engineer or the software engineer would understand to be an irreversible door that they would speak up and let the rest of the team know about that.
00:23:04.738 --> 00:23:13.617
When you would have team meetings in development, was this concept something everybody was aware of that they knew to s- to speak up?
00:23:13.637 --> 00:23:16.317
Or how did you handle situations like that?
00:23:18.089 --> 00:23:22.579
I'd say that it was something that more of my product management team and the...
00:23:22.609 --> 00:23:26.615
I also had a UX team was inculcated with.
00:23:27.224 --> 00:23:36.654
But there were certain leaders in the engineering organization, for example, who had a ton of experience building these kinds of systems.
00:23:37.285 --> 00:23:46.775
And so they were very good about raising their hand and saying,"Hey, you guys are cutting your nose off in spite of your face," types of situations.
00:23:46.865 --> 00:23:52.944
And culturally that initially caused a lot of, animosity, right?
00:23:52.964 --> 00:24:03.394
Because like,"Oh, engineering's just trying to block what we're doing." And you know, we kind of had to shift some mindsets as well to say,"No they're doing their job.
00:24:03.404 --> 00:24:08.164
They're looking at technical feasibility of this overall idea.
00:24:08.634 --> 00:24:32.744
And, you know, we're, we're responsible for value and viability of the overall solution." So we need to, listen to what they're saying, collaborate with them to then be able to come together to form, the right outcome for the customer, the right outcome for our profit margins, you know, for our, our capabilities that we want to ship for customers.
00:24:33.275 --> 00:24:36.174
And so, that I think, you're right.
00:24:36.234 --> 00:24:43.724
There's, um, there's a tension there potentially and I think, it definitely can lead to conflict.
00:24:44.565 --> 00:24:57.404
And so maybe what we could have done better earlier on, frankly, was having some of those shared values for the project m- maybe within the context of this more innovative project.
00:24:57.875 --> 00:25:05.345
And so I think that's something that others might think about when you're spinning up a project like this is how do we want to operate differently?
00:25:05.734 --> 00:25:07.125
What are our values?
00:25:07.494 --> 00:25:12.775
And how do we think about these different decision points in the overall process?
00:25:15.275 --> 00:25:21.753
So considering that potential tension and conflict which is normal, right?
00:25:21.940 --> 00:25:22.200
Mhm
00:25:23.064 --> 00:25:28.651
And we're Talking, we started off with prototypes and how it's so easy and so fast to make them.
00:25:29.070 --> 00:25:29.530
Yeah
00:25:29.911 --> 00:25:34.881
Are we getting to a place where we can use prototypes?
00:25:34.881 --> 00:25:41.201
We primarily think of prototypes for use with the customer or to solve a particular engineering question.
00:25:42.230 --> 00:25:56.861
Do you think with the speed of how we can prototype that it will be able to be used to address some of that tension and the conflict with decisions within the teams, or that's not really a place that you're seeing it fit well?
00:25:59.588 --> 00:26:09.179
Honestly, I think where the industry is right now, it's in the process of trying to sort out How does this new world work?
00:26:09.328 --> 00:26:27.009
Because if an engineer can develop a solution just, you know, or a prototype, just as easily as a product manager or, or a designer, or frankly, it could be, like, an operations person, or it could just be somebody in tech support who's like,"Hey, I had an idea.
00:26:27.009 --> 00:26:28.278
Like, could we fix it this way?
00:26:28.769 --> 00:26:30.509
Or could we create this new thing?
00:26:30.509 --> 00:26:38.739
And hey, I, I used, you know, my prototyping tools that I have available to me at my company," or, you know, or, or what have you.
00:26:39.989 --> 00:26:45.199
Who makes the decision at the end of the day on, like, do we proceed with this particular thing?
00:26:46.088 --> 00:26:57.439
It occurs to me there was an episode of Lenny's podcast, six months ago with one of the product leaders, a growth product manager at, at Anthropic.
00:26:57.449 --> 00:27:00.409
And so, you know, they're in the maelstrom of all of this.
00:27:01.219 --> 00:27:13.038
Who's building and who's not, and how do they, you know, how do they get super fast delivery but, you know, obviously protect protect the business at the same time?
00:27:13.558 --> 00:27:17.999
And they have a couple of ideas that I thought, you know, might be valuable to share here.
00:27:17.999 --> 00:27:42.919
One was as a product manager, if you can find engineers who have good product taste, then leverage them, and leverage them in a certain way, which is give them perhaps the freedom to go and develop something that product management may not have the time to do the sort of the upstream validation work on.
00:27:43.358 --> 00:27:52.459
I think their framework is if you can build it within two weeks or so one sprint, then an engineer can take it from cradle to grave.
00:27:53.068 --> 00:28:08.028
And so they've tapped into these engineers that have what they consider good product taste to be able to then accelerate and leverage the whole the whole team in a different way than they could before.
00:28:08.699 --> 00:28:14.489
So I think there's this idea bubbling around right now of everybody's a builder now.
00:28:15.459 --> 00:28:23.719
And so it's like different shapes of what does a builder look like in a company that's evolving right now as we're having this conversation.
00:28:23.719 --> 00:28:24.689
It's a hot topic.
00:28:25.823 --> 00:28:32.292
Because if you scale this out also into other types of functions, like we're talking about the building stuff.
00:28:33.042 --> 00:28:40.752
But let's say you have an engineer who built a feature or whatever, but maybe they're interested in business models.
00:28:41.623 --> 00:28:53.742
They could work with Claude or GPT or an open weight model or what have you and say,"Hey, like, how could we model this business differently?" And I have...
00:28:53.813 --> 00:28:59.782
Maybe they have access to the the data lake to be able to do some deeper analysis.
00:29:00.143 --> 00:29:08.613
And, they're not maybe experts in business models, but they have access to the data that a product manager, maybe should have access to but doesn't.
00:29:09.063 --> 00:29:13.823
But they could, they could spin up ideas that then bubble out.
00:29:14.492 --> 00:29:29.313
You could have it come from other parts of organizations like,"Hey, what about a engineer creating marketing copy?" Which in the past you would go like,"Egads no.""No engineer can write marketing copy." Like that was like a rule.
00:29:29.613 --> 00:29:40.893
But now with an assistant that can craft, value propositions and, use standard frameworks and stuff to do that, maybe they come up with really good, good ones.
00:29:40.932 --> 00:29:45.113
I've had engineering partners in the past who are great at th-that kind of stuff.
00:29:45.276 --> 00:29:56.066
with the advent of these other tools maybe the boundaries of what work is within organizations has become-- is becoming more and more gray.
00:29:56.675 --> 00:30:08.006
And I don't think it will go totally gray because obviously you need accountability and you need companies to have structure in ways that make sense.
00:30:08.630 --> 00:30:30.140
But it is like a de- democratization of knowledge and information, which then leads to each person within a company being able to take on the functions of other specialists, specialty roles that then could augment or help them, help that other role.
00:30:30.230 --> 00:30:34.779
Like a product marketer helping engineering, but you never know.
00:30:34.849 --> 00:30:40.859
This is where we're really at the frontier of this stuff and it's it's pretty fascinating to see where it's gonna go.
00:30:42.188 --> 00:30:44.718
Yeah, all the lines are getting blurred.
00:30:44.948 --> 00:30:45.367
Yeah.
00:30:45.984 --> 00:30:49.375
one other quick thing from that Moonshots podcast.
00:30:50.125 --> 00:30:55.884
The, I think it was last, maybe a week or two ago, they were providing this idea of, hey, if...
00:30:55.980 --> 00:31:16.299
I think it was one of the guys that was, like, a venture, maybe a venture investor, and he said,"Hey, if I go to my portfolio company and I say, look, thinking about the fact that you now have 100 new employees, quote unquote for free in a sense through these AI agents, what would you use them for?
00:31:16.380 --> 00:31:24.700
Like, how would you actually take advantage of that?" So the, they're selling this message of exponential growth and exponentiality.
00:31:25.210 --> 00:31:35.375
But that's a really interesting way to boil it down and just say hey, if you had 1,000 agents that you could take advantage of, how would you do that with your...
00:31:35.405 --> 00:31:38.645
If you're a solo operator or even if you're a 50-person company.
00:31:39.125 --> 00:31:44.385
That's where it gets hard to think about how do you orchestrate all that?
00:31:44.875 --> 00:31:45.766
What would you work on?
00:31:46.036 --> 00:31:47.226
How do you dashboard it all?
00:31:47.276 --> 00:31:50.097
There's so many questions that pop up in my mind.
00:31:51.542 --> 00:32:10.692
So for the solo engineer working in industry, be-- software, mechanical, electrical, chemical all of these things, there's more opportunity with all of these fast prototyping machines and using them to fill the knowledge gaps to try new things.
00:32:11.300 --> 00:32:11.621
Mhm
00:32:11.722 --> 00:32:23.992
What is a decision framework or advice about decision-making, about moving forward that you would leave with somebody for their Monday morning?
00:32:26.373 --> 00:32:31.230
there's something that we implemented That we called the opportunity brief.
00:32:31.780 --> 00:32:35.451
And an opportunity brief was designed.
00:32:35.461 --> 00:32:38.780
It got it took on a life of its own unexpectedly.
00:32:38.780 --> 00:32:48.351
But the idea came from a blog post from Marty Cagan who's a product kind of product manager guru out of the Bay Area.
00:32:48.871 --> 00:32:51.221
He's written several books which are also very good.
00:32:51.361 --> 00:33:06.750
And the opportunity framework that he has is a series of 10 questions, and it comes down to"Hey who's the market for this thing?" It's the basic stuff that you learn, but it's one of those, oh yeah, the it's so simple we should actually use it ideas.
00:33:06.849 --> 00:33:08.603
But it's what is the product?
00:33:08.603 --> 00:33:09.772
Who's it for?
00:33:09.792 --> 00:33:10.893
How big is the market?
00:33:10.903 --> 00:33:12.752
Is the market actually addressable?
00:33:13.222 --> 00:33:14.353
how large is the market?
00:33:14.353 --> 00:33:20.633
And so it goes through a series of these 10 axes that you can think about and try to run through.
00:33:21.403 --> 00:33:27.076
That's a good one just to give you the back of the napkin, is this thing worth pursuing?
00:33:27.096 --> 00:33:30.936
And I'm thinking more in terms of a product or maybe a feature area.
00:33:31.527 --> 00:33:34.936
But but I would start with the opportunity concept,
00:33:35.007 --> 00:33:40.356
Yeah, that opportunity tracker seems like a good checklist to think through.
00:33:41.057 --> 00:33:47.626
And then I will include links to all of the books that you've mentioned on the podcast blog so people can get to it easily.
00:33:48.989 --> 00:33:49.568
Perfect
00:33:49.997 --> 00:33:56.997
Ashley, I Really appreciate you coming on the show and talking about how things are speeding up and how we make decisions.
00:33:57.190 --> 00:34:07.580
I appreciate the opportunity and, uh, hopefully this is helpful to folks scratching their heads on what's going on right now in the industry and in the world in general.
00:34:08.704 --> 00:34:12.434
If anyone wanted to reach out to you, what is the best way they could do it?
00:34:14.081 --> 00:34:14.490
Yeah.
00:34:14.541 --> 00:34:15.331
Two things.
00:34:15.331 --> 00:34:18.501
One is I have my substack called Beyond the Prototype.
00:34:19.010 --> 00:34:21.951
And then, feel free to connect with me on LinkedIn.
00:34:22.001 --> 00:34:26.061
And I think you'll probably link those in the, the notes, show notes as well
00:34:26.494 --> 00:34:29.684
Yes, I'll link to your contact information in the show notes.
00:34:31.105 --> 00:34:31.625
Sounds great.
00:34:31.684 --> 00:34:32.155
Thanks, Dianna
00:34:36.050 --> 00:34:38.139
That concludes the interview with Ashley.
00:34:39.170 --> 00:34:43.909
Ashley and I talked about what changes when building becomes fast and inexpensive.
00:34:44.679 --> 00:34:51.219
The opportunity is that we can test more ideas and learn more quickly, but the discipline still matters.
00:34:51.690 --> 00:35:00.170
Before adding another feature or polishing another prototype, we need to ask whether we have evidence that the problem is worth solving.
00:35:01.380 --> 00:35:06.590
That may be one reason we jump when the next build feels fast and inexpensive.
00:35:06.900 --> 00:35:13.230
It can feel like progress before we have clarified the problem, the customer, or the evidence we need.
00:35:14.409 --> 00:35:23.539
We also talked about using AI as more than a builder, as a thought partner, a critic, or a structured set of perspectives.
00:35:24.340 --> 00:35:36.150
And we explored the difference between two-way door decisions, which can be tested and reversed, and one-way door decisions, which need more deliberate thinking because they create longer term commitments.
00:35:37.960 --> 00:35:44.659
So here's your Monday morning move: choose one idea, feature, or project you're considering.
00:35:45.579 --> 00:35:54.550
Write down who it's for, what problem it addresses, what evidence would show that it's valuable, and whether your next decision is reversible.
00:35:55.559 --> 00:36:01.289
If it's a two-way door, design a small test and get it in front of someone who can give you real feedback.
00:36:03.210 --> 00:36:07.679
You can find Ashley through his Substack, Beyond the Prototype, and on LinkedIn.
00:36:08.239 --> 00:36:12.050
I'll include those links along with books that Ashley recommends.
00:36:16.153 --> 00:36:17.833
I'll include those links.
00:36:19.184 --> 00:36:23.653
You can find Ashley through his Substack, Beyond the Prototype, and on LinkedIn.
00:36:25.184 --> 00:36:30.813
You can find Ashley through his Substack, Beyond the Prototype, and you can find him on LinkedIn.
00:36:31.804 --> 00:36:39.804
I'll include those links along with the books and podcast recommendations that Ashley wanted to give you in the show notes.
00:36:41.273 --> 00:36:55.114
If this conversation helped you think differently about speed, decision-making, or the build trap, subscribe to Quality During Design and share this episode with someone who is moving quickly on an uncertain product or project.
00:36:56.224 --> 00:37:03.603
You can also find more reflections in the monthly digest at newsletter.deeneyenterprises.com.
00:37:05.003 --> 00:37:07.514
This has been a production of Deeney Enterprises.
00:37:07.974 --> 00:37:08.833
Thanks for listening.